Studies in African Linguistics Volume 14, Number 3, December 1983

ERGATIVITY AND THE ACTIVE-STATIVE TYPOLOGY IN LOMA

Noel Rude University of Oregon

Ergativity would seem to be non-existant or at least quite rare in Africa. This lack, however, may be related to an other continent-wide areal phenomenon: there is a paucity of morphological NP case marking according to either erga tive or accusative typologies. It is thus possible that other more subtle attributes of the ergative organization of syntax are what should be sought in Africa. For exam ple, in the Mande languages, as also in Celtic, phonologi cal decay has produced a series of word initial consonant alternations. In Celtic these have come to function as part of a nominative-accusative case marking strategy. The situation is quite similar in Mande, but as this paper details for Lorna, the noun case system is ergative-absol ute. And, accordingly, the pronoun system has active stative characteristics.

O. Introduction The Mande languages are famous for indigenous syllabaries and peculiar consonant mutations (see Welmers [l97lb]). Loma,l the language of this in vestigation, has both a syllabary and consonant mutation. Lorna's consonant mutation has received repeated mention in the literature, e.g. Eberl-Elber [1937], Hintze [1948], Manessy [1964], Meeussen [1965], Welmers [197lb], Bird [1971], Hyman [1973]. Dwyer [1974]. There is also a grammar by Sadler [1951] and a brief description of the language by Heydorn [1971]. Although the lin guistic features that are dealt with in this paper have been documented else where, no one as yet, to my knowledge, has couched his description in the ter minology of ergative and active-stative typologies. 2

IThe data for this paper was gathered in the autumn of 1976. It was kind ly provided by Sewalla Guseh, then a student at the University of Oregon. Mr. Guseh was a native speaker of Lorna from Zenalomai, Liberia. 2Heine and Reh [1982J note that "although the IOOO-odd African languages display a remarkable extent of structural variation, there are certain struc- 266 Studies in African Linguistics 14(3), 1983

1. Ergativity 1.1. Consonant Mutation. There is a phonological alternation in Lorna that happens only word initially and only in well defined syntactic environments.

Th~ alternations are as follows (~ and y appear before unrounded vowels, w before rounded vowels):

p, b -+ ~ w t -+ v t, d -+ 5 -+ z k, 9 -+ Y w z -+ y kp -+ b

1. 2. Syntactic environment for consonant mutation. 1.2.1. The noun J2hrase. Word initial consonant mutation never occurs initially in a noun phrase. In the following examples, the citation forms of nouns in (a) occur without consonant mutation, while consonant mutation is illustrated in (b) where a morpheme precedes. 3

(1) a. p€l€i 'the house' b. ga p€l€i 'our house' (2) a. bu Ii 'the goat' b. ga wu Ii 'our goat' (3) a. t i 'the work' b. ga Ii 'our work' (4) a. d;:,i 'the wine' b. ga I;:, i 'our wine' (5) a. ki 'the key' b. ga yi 'our key'

tures that do not seem to occur in Africa. Thus, to our knowledge, an erga tive case or a numeral classifier system has not been discovered so far." For the ergative typology see Comrie [1978) and Dixon (1979); for the active stative typology see Klimov [1974). 3The mutated b is described by Sadler [1951] as a labio-dental stop, as opposed to v which also occurs in the language. ~~en I collected this data seven years ago, my impression was of a voiced bilabial fricative. But what- ever the sound, it is in this paper represented by 0 Tones are not indi- cated in this paper unless relevant to the argument being made. Most Lorna words have high tone inherent in the first syllable. There is no tone con trast, for example, between ~a and ba as in zunui pa 'for the man' and ba 'for him', even though the high tone in ba contrasts with the low tone in ba 'for me'. For common segmentally identical morphemes like e 'he/ she/it' and e 'you', only the morpheme with low tone will be marked in this paper. The data from other Mande languages was gleaned from various sources in which tones are not distinguished in a consistent manner. Ergativity in Lama 267

(6) a. guli 'the tree' b. ga wu Ii 'our tree' (7) a. kpug i 'the door' b. ga bugi 'our door' (8) a. to I i 'the sun' b. ga vo I i 'our sun' (9) a. siyi 'the cloth' b. ga z i y i 'our cloth' (10) a. ziyi 'the hole' b. ga y i Y i 'our hole'

In (11), consonant mutation has occurred word initially everywhere, but not phrase initially.

(ll) ta yunu 0ai pa 'for their good man' their man good-the for

The following two examples restore the non-mutated forms of zunu 'man' and ba 'for'. The non-mutated form for 'good' is pa The 3rd person singular pronoun in (13) is merely the high tone.

(12) zunu 0a i 0a 'for the good man' man good-the for (13) ba 'for him/her/it'

1.2.2. The verb phrase. 1.2.2.1. Intransitive. Consonant mutation occurs in the initial consonant of an intransitive verb when its subject precedes. Compare the following. The toneme in (15) permits the verb to be phrase initial and thus preserve the non-mutated form of its initial consonant.

(14 ) zunui !3a I a 0t: 'the man is big' man-the big ASP (15) bala 'he/she/it is big' big ASP

The intransitive verbs di 'go', tt: 'climb', and si I i 'arrive' occur in the following examples. A preceding subject in every case is sufficient to bring about consonant mutation. In (19) the verb is preceded by an auxiliary ending in a nasal consonant and still the initial consonant of the verb mu tates.

(16) zunui Ii zu 'the man is going' man-the go ASP (17) bu Ii IF: ne 'the goat climbed' goat-the climb ASP 268 studies in African Linguistics 14(3), 1983

(18) see palai z iii ni 'the large elephant arrived' elephant big-the arrive ASP

(19) zunui y,~ f) f:3ala ne 'the man used to be big' man-the ASP big ASP

If the predicate is not a verb, but instead is initiated by a NP, it should be noted that no mutation occurs between the subject and predicate. The postposition in (21) is initial in the predicate due to the pronominal to neme. Compare the non-mutated pEIEi in (20) with pEIEi in (lb) above. In (22) neither the copula ga undergoes consonant mutation nor the predicate noun zunu .

(20) zunui Pt~ I (' i wu 'the man is in the house' man-the house-the in (21) zunui bU 'the man is in it' man-the in , (22 ) ta ga zunu 'they are men' they be man

1.2.2.2. Transitive. Mutation affects the initial consonant of a transitive verb only when a direct object precedes, as in (23) and (24) below. The tran sitive verb tE 'lift, raise' is the same morpheme as the intransitive 'climb' in (17) above. The verb is made transitive by the presence of both S and 0, as can be seen by comparing (23) with (17) above. Note that the ini tial consonant of the direct object does not mutate even though the subject precedes. Compare the non-mutated kJti 'stone' in (23) with ga wJti 'our stone' .

(23) zunui kJt i IE ne 'the man lifted the stone' man-the stone-the lift ASP (24) zunu i bu Ii f:3aa ne 'the man killed the goat' man-the goat-the kill ASP

When the direct object is realized as only a tone then the subject direct ly precedes a transitive verb. Unlike the subject preceding an intransitive verb, no mutation ever occurs when an agentive subject immediately precedes a transitive verb. The transitive verb in (25) is distinguished from the in transitive one in (17) above solely by not having undergone consonant muta tion. Although both verbs below have inherent high tone, the preservation of Ergativity in Lorna 269

the high tone in (25) and (26) indicates an anaphoric 3rd person singular di rect object.

(25 ) zunui ts ne 'the man lifted it' man-the lift ASP , (26) zunui paa ne 'the man killed it' man-the kill ASP

Thus Lorna clearly evidences ergativity. The environment for mutation of the initial consonant of the verb is an immediately preceding absolutive NP. And this Lorna absolutive case includes, in the classic sense, both direct ob jects of transitive verbs and subjects of intransitive verbs. When an ana phoric direct object is realized as zero, the verb is immediately preceded by an ergative NP. And in this environment the initial consonant of the verb never mutates.

1.3. Historical development of consonant alternations. 1.3.1. Phonology. The historical phonology of the Southwestern Mande lan guages has been discussed by Hyman [1973] and by Dwyer [1974]. Historically, consonant mutation was blocked in Lorna by a preceding nasal consonant. This nasal was later completely lost in Lorna. But in Mende it still survives be fore voiced consonants, as seen in the following: 4

Lorna Mende (27) daba ndamba 'crocodile' (28) kpadE kpandE 'gun' (29) ga nga 'I'

Before voiceless consonants the nasal was lost in both Mende and Lorna. But it does survive in Kpelle, although the consonant that it shielded from weakening in Lorna and Mende is in Kpelle lost completely. All clusters N+C are simpli fied to N in Kpelle, as can be seen below. The correspondence set in (31) suggests the reconstruction *kontu with the cluster nt , this on analogy with correspondence set (30).

4Mende examples are from Innes [1967,1969], Kpelle from Welmers [1962, 1969]. See Manessy [1964] for a description of the blocking of consonant len ition by word final *') • 270 Studies in African Linguistics 14(3), 1983

Lorna Mende Kpelle (30) kpadc: kpandc: kpana 'gun' (31) k:::>t u k:Jt u konu 'stone'

The only permissible word final consonant in Proto-Southwestern Mande was

~. This consonant was lost in Lorna but survived in Kpelle, as is illustrat ed below.

Lorna Kpelle (32) b bn 'law' (33) kpa I a kpalaf) , farm'

Word final f), however, does have a reflex in Lorna. When a vowel is suffixed to such a word, the surfaces as g. The following examples contrast nouns with and without this final nasal. The Lorna nouns pc:lc: 'house' and zala 'lion' did not end in a nasal in the proto-language. The word for 'house' in Kpelle, for example, is pc:rc:, not *Pfrfll. The nouns kpala , farm' and daba 'crocodile' have cognates in other Mande languages with final nasal, and in examples (36) and (37) this nasal surfaces in Lorna when a vowel is suffixed, as in (36b) and (3 7b) .

(34) a. pc: I f i 'house' b. pc:lc:i 'the house' (35) a. zala 'lion' b. zalai 'the lion' (36) a. kpala 'farm' b. kpa lag i 'the farm' (37) a. daba 'crocodile' b. dabagi 'the crocodile'

The point of all this, as the following examples show, is that this nasal blocks the otherwise expected lenition in the initial consonant of a follow ing word. In example (39), lenition of the initial consonant was blocked, as is evident from example (36), by an immediately preceding nasal consonant.

(38) pc:l, 'the good house' (39) kpala pai 'the good farm'

Lenition of the first consonant of a verb is also blocked when a direct object that once ended in a nasal precedes, as the following examples show. The initial consonant of paa 'kill' does not weaken after daba 'crocodile', because that noun ended in a nasal in the parent language, as example (37) Ergativity in Lorna 271

above indicates.

(40) za I a j3aa 'kill a lion!' (41) daba paa 'kill a crocodile!'

This phenomenon also affects the aspectua1 morpheme su which normally mutates to zu after verbs, but not after all verbs. The same verbs that block the lenition of su also regularly occur with final 9 when the as pect marker a is suffixed. Examples (42) and (43) below show verb stems that do not block lenition, while the verb stems in (44) and (45) do. , (42) a. ga Ii zu ' I am going' b. ge Ii a 'I have gone' I go ASP I go ASP , (43) a. ga kc: zu ' I am doing it' b. ge kc: a ' I have done it' I do ASP I do ASP , , (44) a. ga bale su ' I am sweeping' b. ge baleg a ' I have swept' I sweep ASP I sweep , (45) a. ga b;) su , I am helping' b. ge 6;)g a , I have helped' I help ASP I help ASP

Kpe11e, which preserves word final *~, preserves the final 8 in those verbs that block lenition in Lorna. The Kpe11e cognates of Lorna I i 'go' and kc: 'do' are I i 'go' and kc: 'do', both without final 8. But Kpe11e has

kp;)~ 'help' where Lorna has kp;) 'help'. The verb for 'sweep' in Lorna also occurs as a noun, viz. kpa Ie/ kpa leg i 'broom/ the broom'. And so as a noun the reflex of *~, the 9 , also surfaces before a vowel suffix.

1.3.2. Morphology. The two major syntactic environments that block consonant lenition in Lorna are (1) the initial position in a NP and (2) the initial po sition in a transitive verb when no direct object NP precedes. There is ample comparative evidence that historically both environments were marked by a na sal prefix that blocked consonant lenition.

1.3.2.1. Lenition blocked in verb. In Lorna, all that survives of the 1st and 3rd person singular direct object pronouns is non-lenition of the initial con sonant of the verb and a low or high tone on the first syllable of the verb, as shown below:

(46) e buli luli ni 'he called the goat' he goat-the call ASP 272 Studies in African Linguistics 14(3), 1983

(47) e t uI i ni 'he called me' he call ASP (48) e t uI i ni 'he called him/her/it' he call ASP

The prefixed pronouns that blocked lenition in the Lorna verb are reconstructed as *N- 'me' and *'N- 'him/her / it' . There has been a reversal of tones in Lorna, for which see Welmers [197laj.5 As one would predict (see again ex amples (27-29), the nasal prefix survives before voiced consonants in Mende, as can be seen in the following examples. In some of them the transitive ver sus intransitive use of the verb illustrates the presence of the direct object prefix in (b) , in other cases, e.g. (49) and (52), a direct object NP precedes and thus lenition occurs in (a) .

(49) a. nCJi waa 'kill him/her' b. paa 'kill it' (50) a. 12 'climb, rise' b. t2 ' lift it, raise it' (51) a. gutu 'be short' b. kutu 'shorten it' (52) a. ndJ gbJ 8 'drink palm wine' b. kpJ8 'drink it' (53) a. be 'dry' (intransitive) b. mbe 'dry it' (54) a. la 'lie down' b. nda 'lay it down' (55) a. yei 'descend' b. njei 'lower it' (56) a. wua 'enter' b. ngua 'insert it'

Historically, initial consonant lenition in the verb was blocked by the same phonetic environment as elsewhere; it was blocked by a prece.ding nasal conso nant. The fact that this consonant was a grammatical morpheme explains the syntactic behavior of ve.rb initial consonant lenition in modern Lorna.

1.3.2.2. Lenition blocked in noun. Greenberg [1977] hypothesizes that the Niger-Congo noun class markers originated as definite markers, whether prefix or suffix. The process involves three stages: definite markers > referen tiality markers > substantive markers. All three stages of this develop-

5Welmers proposes ;'~ as the pbonetic form of the third person object prefix, but others-tlanessy [1964], Meeussen [1965], Hyman [1973J, Dwyer [1974]-argue for "N, ,,,hich is to be preferred, both in view of what regu larly blocks lenition in Lorna and from the point of view of a possible cog nate in Niger-Congo. Ergativity in Loma 273

ment are observable in various of the Niger-Congo languages for both prefixal and suffixal systems. Greenberg [1977:97] observes that "almost every branch of Niger-Congo, except of course Mande, has some languages which are simultan eously prefixing and suffixing." In the West Atlantic languages, Greenberg notes, the prefixes have advanced to the last stage, and in some languages of the group they gave been reduced to noun initial consonant alternations. But suffixed articles have been innovated, and in various of these languages the different stages of their development are visible. Greenberg also notes how the Southwestern Mande noun suffix -i is in the first stage in Lorna but more advanced in Mende. All this provides a very helpful context for viewing the historical devel

opment of the Lorna noun initial consonant alternations. '~elmers [197lb] hy pothesizes that the 3rd person pronominal now realized in Lorna simply by its blocking consonant lenition in the verb is also responsible for blocking con sonant lenition initially in the noun. And thus the Mande languages do in deed, along with the other branches of Niger-Congo, evidence both prefixing and suffixing. Lorna d;) 'palm wine' is made definite by suffixing -i : d;) i 'the palm wine'. The lenition of the d, ,,,hich should have become I , was blocked by a nasal prefix that has been lost in Lorna. This nasal prefix was an obligatory final stage substantive marker. It has survived in Mende before nouns with initial voiced consonants, as the following examples indi cate.

Lorna Mende (57) bala mbala 'sheep' (58) d;) nd;) 'palm wine' (59) z i e nja 'water' (60) gulu ngulu 'tree' Nouns are derived from verbs by the nasal prefix. Consider the following ex amples from Mende. Remember that in Mende a nasal consonant Dlus voiceless obstruent simplifies to just the voiceless obstruent but with lenition being blocked, as seen in (6lb). The nasal survives, though, before voiced conso nants, as in (62b) and (63b). 274 Studies in African Linguistics 14(3), 1983

(61) a. wa 'come' b. pa 'a coming' (62)a. Ii 'go' b. nd i 'a going' (63) a. yenge 'work' (verb) b. ngenge 'work' (noun)

In Kpelle the prefixed nasal was not generalized to mark all nouns. It still functions as a definite marker, but redundantly along with the suffix -i The presence of the nasal prefix has had the opposite effect in Kpelle as in Lorna and Mende: it has provided the environment for consonant lenition. Its effect was to voice a voiceless consonant and to bring about the total loss of a voiced one. The nasal itself survives only where it was prefixed to a voiced consonant. The original 10'01 tone of this prefix survives in Kpelle, where definite nouns begin with a low tone. The suffix -i has been lost af ter nouns ending in f"), as can be seen in (65), (67), and (74).

(64 ) a. psn: 'house' b. bsn: i 'the house' (65) a. bl"") 'law' b. d :>r) 'the law' (66 ) a. kJ:) 'log' b. gJJi 'the log' (67) a. kpalaC) 'farm' b. gbalan 'the farm' (68) a. boa 'knife' b. moai 'the knife' (69) a. luu 'fog' b. nuui 'the fog' (70) a. ya 'water' b. nyai 'the water'

(71) a. y i I a 'dog' b. I"") i I a i 'the dog' (72 ) a. wuru 'tree' b. flur i i 'the tree' (73) a. fena 'mushroom' b. venai 'the mushroom'

(74) a. SU) 'thing' b. ZEI) 'the thing'

In the Mande languages, unlike in the rest of Niger-Congo, there is no ev idence for noun classification. However, the Mande substantive marking prefix may be cognate to a Niger-Congo noun class marker. The Mande prefix was a homorganic nasal just as was the Bantu class 9 prefix. Further, as Givan [1971] shows, the Bantu class 9/10 probably originally marked animates, class 1/2 having arisen later to mark the subcategory human. If Mande were to have generalized one of the Proto-Niger-Congo class markers to all nouns, the ani macy marker would obviously have been the best candidate. And, according to Greenberg [1977], this would also have been a definite marker, as it still is Ergativity in Lama 275

in Kpelle. The Mande substantival prefix and 3rd person object pronoun are phonetically identical, and Welmers [197lb] assumes the object pronoun to pre serve the original function of the morpheme. 5 But would this not indicate that the first nouns to be marked definite were objects? To believe so one would need more Niger-Congo evidence, for to my knowledge no one has yet re constructed noun case marking in any of Niger-Congo. According to Welmers [197laj, morphemes of class 1 and 9 in both Bantu and the closely related Tiv manifest low tone, while for all other noun classes the tone is high. Signif icantly, the Mande nasal prefix is also reconstructed with low tone.

1.3.3. Emergent ergativity. Synchronically, Lorna clearly displays an erga tive syntax. The subject of an intransitive verb and the direct object to gether provide the syntactic motivation for verb initial consonant lenition. Although no morphemes are involved, the syntactic organization reflected by the phonology is ergative-absolutive. The ergative NP sits outside the phras al unit which comprises the absolutive NP and the verb, as illustrated in (75). The # shows where consonant lenition is blocked.

(75) (IIERGATIVE NP) [#(ABSOLUTIVE NP) VERB]

Here is an ergative-absolutive system that has arisen quite by accident. There was no reanalyzed passive with marked agent, nor nominalization 'vith genitive agent. The first prerequisite in the development of the Lorna system was the rigid SOV word order with, of course, SV for intransitive clauses. This allows for a syntactic organization as follows:

6The prefix *N- evidently also served as a kind of resumptive subject pronoun in Lorna. Compare ~a 'come' with ba 'come' in the examples below. Consonant mutation occurs as expected in (a), but not in (b) where the verb is preceded by a relative clause. The verb koa does not mutate to woa be cause it is transitive. (1) zunui ,3a ne 'the man came' man-the come ASP (2) zunui 98 koa ne ba ne 'the man I know came' man-the I him-know ASP come ASP 276 Studies in African Linguistics 14(3), 1983

(76) a. Intransitive S V

b. Transitive S o V ERGATIVE ABSOLUTIVE VERB

The second prerequisite in the development of the Lorna system was the fact that *'N- was both a substantive marker as well as direct object pronoun, as diagrammed below. , (77) a. Intransitive N-S V , , h. Transitive 1) with d. o. noun N-S N-O V , , 2) anaphoric d. o. N-S N-V

When the Lorna consonant lenition occurred, *N- provided a syntactic environ ment for its non-occurrence. Then, after N+C was everywhere simplified to C, only the famous consonant alternations were left as syntactic indicators of transitivity. 7 Another factor in the development of the system is the Mande characteristic of coding transitivity as NP NP V. With a transitive verb the NP in the sequence NP V must be interpreted as a patient. Bird and Shop en [1977] descri.be this system for Maninka, where it is extremely productive. Consider, for example, the following Haninka sentences.

(78) an be 50go dumu 'we are eating meat' we ASP meat eat

------.----- 7lt is peculiar that the world's most outstanding examples of grammatical ized consonant alternations should hug the western extremities of the Old World. Celtic, Berber, West Atlantic and Mande are all famous for consonant alternations tied to syntax/semantics. The same observation cannot be made, for example, with regard to the eastern extremities of the Eastern Hemisphere. Northwest Semitic, which includes the Phoenician carried west by sea, is also famous for its syntactically linked spirantization of consonants. Even in Ro mance it is in the West that syntactically linked phonological processes de veloped; spirantization in Spanish, liaison in French. Even Germanic, that other western extension of Indo-European, is famous for spirantization, though there is no record of it ever having heen grammatically linked. The Hopper [1977] analysis of Proto-lndo-European makes Grimm's Law look exactly like the Northwest Semitic spirantization. Of course similar phenomena have arisen spontaneously allover the globe. There may have been more coinci dence than Phoenician in the Niger-Congo, Berber, and Indo-European hlest. Nevertheless, in no other region of the world has grammatically linked conso nant alternation been carried to such extremes over such a large area. Ergativity in Lorna 277

(79) an be dumu 'we are being eaten' we ASP eat

In Lorna, when a single NP stands before a mutated transitive verb, it also must be the patient, as the following illustrates.

(80) e k::> t i 12 zu 'he is lifting the stone' he stone-the rise ASP (81) e t2 zu 'he is lifting it' he rise ASP (82) e Ie: zu 'he/she/it is rising' he rise ASP

2. Active-Stative Typology

2.1. Pronoun system. There are two basic pronoun sets in Lorna that I shall label active and stative. I list them below as pronounced by my informant. Each set has two pronouns for 1st person plural which, in the order listed, mark the exclusion and inclusion of the addressee.

ACTIVE STATIVE singular plural singular plural , , , 1st person ga ga, da - ge, de v' , , , 2nd person Ja wa e u 3rd person toa tel !. te

2.2. Function of pronoun sets.

2.2.2. Subject and object. One function of the two sets of Lorna pronouns is to mark subject and object. The "active" set marks subject, the "stative" set object. However, the order of morphemes is rigid and thus these morphological case distinctions are redundant. The following examples illustrate the case function of the two sets of Lorna pronouns. In (85) and (86) the object pro nouns are simply high and low tones in the first syllable of the verb. The stative set also functions as object of postposition, as (87) shows. , (83) ga te '{a zu 'we see them' we them see ASP , (84) ta ge '{a zu 'they see us' they us see ASP 278 studies in African Linguistics 14(3), 1983

, (85) ga ka zu ' I see him/her/it' I him-see ASP (86) toa ka zu 'he/she sees me' he me-see ASP , (87) ga bb su t~ ma , I am telling it to them' I it-tell ASP them to

2.2.2. Aspectual split. The case distinction marked by Lorna pronouns is not maintained in the perfective aspect, as illustrated below. Examples (88) and (89) should be compared with (83) and (84). The pronouns are all from the stative set. , (88) ge t,e ' '{a ne 'we saw them' we them see ASP , (89) te ge ya ne 'they saw us' they us see ASP

Obviously the two tonemes from the stative set cannot mark both subject and , object on the same verb. In the perfective aspect 'I' and e 'he/she/it' are substituted for subject pronouns, as in the following examples: , (90) ge ka ne ' I saw him/her/it' I it-see ASP , (91) e ka ne 'he/she saw me' he me-see ASP

2.2.3. Alienable and inalienable possession. As illustrated in (92) and (93), the active set of pronomina1s functions to show alienable possession, the sta tive set to mark inalienable possession.

(92) ga ~1E lEi 'our house' our house-the , (93) ge 'IE 'our father' our father

For the 1st and 3rd person singular members of the active set na 'my' and na 'hls/her' are substituted as alienable possessors. As seen below, they block consonant mutation. For this reason Manessy [1964] suggests that they once ended in nasal consonants.

(94) na PE lEi 'my house' Ergativity in Loma 279

, (95) na pEIEi 'his/her house' (96 ) kE 'my father' (97) kE 'his/her father'

2.2.4. Active versus stative subject. It is their function as subject of intransitive verbs that provides the justification for calling the Lorna pro noun sets active and stative. The following four sentences contrast an ac tive/agentive versus a stative subject, both with intransitive predicates. 8

8No negative sentences were elicited during the brief period of my field work on Lorna. But according to Sadler [1951], no matter what the verbal as pect, the subject pronoun of a negated clause is always from the stative set, but with what appears to be vowel harmony with the negative mar~er IE. Sadler gives two forms for the 3rd person singular pronoun of the active set: t6waa , which occurs with the future, and t:5, which occurs with the progres sive. The following examples from Sadler illustrate this contrast. For (b), my informant had the pronoun toa (see example (100) of the text). (a) towaa I ( 'he/she will go' he go (b) t:5 Iii zu 'he/she is going' he go-ASP

(d) ga Iii zu ' I am going' I go-ASP Sadler also has a third set of pronouns for the habitual aspect. Both the habitual and future are marked by the absence of an aspectual suffix. While the future employs pronouns from the active set, the habitual has its own set of pronouns. The following examples have been gleaned from Sadler: , (e) ga Ii ' I will go' I go (f) g~ Ii ' I go (habitually) , I go (g) towaa I r 'he/she will go' he go , (h) ;) Ii 'he/she goes (habitually) , he go There is much more, to be sure, that can be said about the Lorna pronominal system. However, it must suffice for now merely to point out the active stative typology evident in the system. Sadler is a gold mine of data for anyone interested in pursuing the matter further. 280 studies in African Linguistics 14(3), 1983

, (98) ACTIVE SUBJECT: ga Ii zu 'we are going' we go ASP , (99) STATIVE SUBJECT: ge ~ala ~e: 'we are big' we big ASP (100) ACTIVE SUBJECT: toa Ii zu 'he/she is going' he go ASP (101) STATIVE SUBJECT: bala ,Se: 'he/she is big' him-big ASP

The active-stative distinction for intransitive subjects does not hold in the perfective. There intransitive verbs take the same subject pronouns from , the stative set, including g8 'I' and e 'he/she/it', that function as subjects of transitive verhs in the perfective. Compare (102) with (98) and (99). Example (103) has the subject pronoun e. the same as the transitive verb in (91) . , (102) ge Ii n i 'we went' we go ASP , (103) e Ii ni 'he/she went' he go ASP

The contrast between perfective event and inactive state is illustrated by comparing the example below with (101).

(104) e ?ala ne 'he/she grew' he big ASP

Subjects that take pronouns from the stative set are not simply patient, as opposed to agent, but more narrowly patient of state. The patient of change in the following requires a subject pronoun of the active set.

(105) e 10 ne 'he fell' he fall ASP (106) toa 10 zo 'he is falling' he fall ASP

The patient subject of a nominal predicate also selects pronouns from the ac tive set. 9 Example (22) from the text above is repeated here, where ta is

9In Ashanti Twi predicate adjectives require no copula «a) below), loca tive predicates require the copula WJ (b), and nominal predicates the copula Ergativity in Lorna 281

from the active set of pronouns.

(107) t6 ga zunu 'they are men' they be man

Yf (c) . The copula YE is also the verb 'do' and 'make' as in (d) and (c) . Could it be that nominal predicates are somehow more "active" than adjectival? (a) Kof i so 'Kofi is big' Kofi big (b) Kof i W:J ef i e no mu 'Kofi is in the house' Kofi be house the in (c) Kof i Y€ :Js:Jfo:J 'Kofi is a priest' Kofi be priest (d) Kof i YE adwuma no 'Kofi does the work' Kofi do work the (e) Kof i YE a bodoo 'Rofi makes cornbread' Kofi make cornbread 282 studies in African Linguistics 14(3), 1983

REFERENCES

Bird, C. S. 1971. "Observations on initial consonant change in Southwestern Mande." In Chin-\~u Kim and Herbert Stahlke (eds.), Papers in African Linguistics, pp. 153-174. Edmonton, Alberta; Champaign, Illinois: Linguistic Research. Bird, C. S. and T. Shopen. 1979. "Maninka." In Timothy Shopen (ed.), Languages and Their Speakers, pp. 58-111. Cambridge, Mass.: Winthrop Publishers. Comrie, B. 1978. "Ergativity." In Hinfred P. Lehmann (ed.), Syntactic Typo logy, pp. 329-394. Austin: University of Texas Press. Dixon, R. M. IV. 1979. "Ergativity." Language 55:59-138. Dwyer, D. J. 1974. "The historical development of Southwestern Mande conso nants." Studies in African Linguistics 5:59-94" Eberl-Elber, R. 1937. "Der konsonantische Anlautwechse1 in der Sprachen gruppe Gbande-Loma-Mende." Mitteilungen des Ausland Hochschu1e an der Universi t;it Berlin 40: 128-143. Givan, T. 1971. "Some historical changes in the noun-class system of Bantu, their possible causes and wider implications." In Chin-Wu Kim and Her bert Stah1ke (eds.), Papers in African Linguistics, pp. 33-54. Edmonton, Alberta; Champaign, Illinois: Linguistic Research. Greenberg, J. H. 1977. "Niger-Congo noun class markers: prefixes, suffixes, both or neither." In Martin Nould and Thomas J. Hinnebusch (eds.), Pa pers for the Eighth Conference on African Linguistics, pp. 97-104. Studies in African Linguistics, Supplement 7. Heine, B. and Mechthild Reh. 1982. Patterns of Grammaticalization in African Languages. Arbeiten des Ko1ner Universalien Projekts Nr. 47. Ka1n: Institut fur Afrikanistik der Universitat zu Kaln. Heydorn, R. W. 1971. "Grammar of the Lorna language with some K::m;) words in the vocabulary." Afrika und Ubersee 54:77-99. Hintze, F. 1948. "Zum konsonantischen Anlautwechsel in einigen westafrikan ischen Sprachen." Zeitschrift fur Phonetik, Sprachwissenschaft und Kommunikationsforschung 2:164-182, 322-335. Hopper, P. J. 1977. "The typology of the Proto-Indo-European sequential in ventory." ,Journal of Indo-European Studies 5:41-53. Hyman, L. M. 1973. "Notes on the history of SoutllloJestern Mande." Studies in African Linguistics 4:183-196. Innes, G. 1Y67. A Practical Introduction to Mende. London: Luzac aud Com pany. Innes, G. 1969. A Mende-English Dictionary. London: Cambridge University Press. Ergativity in Lorna 283

K1imov, A. 1974. "On the character of languages of active typology." Lin guistics 131:11-25. Manessy, G. 1964. "L'a1ternance consonantique initia1e en manya, kpelle, lorna, bandi et mende." Journal of African Languages 3:162-178. Neeussen, A. E. 1965. "A note on permutation in Kpe11e-Nende." African Language Studies 6:112-116. Sadler, W. 1951. untangled Lorna. Nonrovia, Liberia: Lutheran Nission. We1mers, W. E. 1962. "The phonology of Kpe11e." Journal of African Lan guages 1: 69-93. We1mers, W. E. 1969. "The morphology of Kpel1e nomina1s." Journal of Afri can Languages 8:73-1Q1. We1mers, W. E. 1971a. "The typology of the Proto-Niger-Kordofanian noun class system." In Chin-Hu Kim (ed.), Papers in African Linguistics, pp. 1-16. Edmonton, Alberta; Champaign, Illinois: Linguistic Research. Helmers, W. E. 1971b. "Niger-Congo, Mande." In Thomas A. Sebe.ok (ed.), Linguistics in Sub-Saharan Africa, pp. 113-140. Current Trends in Lin guistics. 7. The Hague; Nouton.